The beginning of the modern approach to artificial intelligence (AI) can be traced back to a summer camp held at Dartmouth University in 1956, which was attended by many scientists who later became prominent in the field of artificial intelligence. Even then, discussions revolved around how machines would become as intelligent as humans within a few generations.

History took a different course, and on several occasions AI development was abandoned (so-called AI winter). Today, we are in a position where the rapid development of language models has opened up many exciting new possibilities. On the other hand, it has also made people worry about what kind of work they will be able to do in the future. What is our future in the age of artificial intelligence? What skills will be needed in the job market?

What can we say about the future?

First, it should be noted that the assessments and opinions written here are based on the best knowledge of early 2025 and the latest developments. No one can predict the exact future.

With the increasing popularity of artificial intelligence, there will be a very high demand in the near future for higher-level experts who can develop, evaluate, manage, and direct AI systems. The IT sector has been a very successful employer for many years and constitutes a significant part of Estonia's exports. In 2023, Estonia's ICT sector exported services worth 3.6 billion euros, accounting for 12 percent of total exports of goods and services. In the third quarter of 2024, the average gross salary in the information and communication field in Estonia was 3,506 euros, or about 1.8 times higher than the Estonian average salary.

Let's look at more figures. According to Statistics Estonia, the share of companies using machine learning in Estonia was 1.6% in 2023, and already 3.8% in 2024. At least one artificial intelligence technology was used by 5.1% of companies in 2023, and already 13.9% in 2024. For comparison: according to Eurostat, the latter share in Denmark was 27.6% in 2024, in Sweden 25.1%, in Finland 24.4%, with the European Union average at 13.5%. Although we are slightly above the European Union average, it's clear that we could aim significantly higher.

However, there is no reason to believe that by the end of this year we will reach the 30% threshold. Such a leap requires a very significant paradigm shift for many companies. This change can be brought about by AI engineers who understand how to develop and manage machine learning or neural network processes that fall under artificial intelligence.

Changing role of the developer

The profession of developer or programmer is changing. Artificial intelligence is very good at writing code that it has seen before, or code that needs to be written in a similar form in many applications. It is good at writing very specifically defined solutions where instructions need to be provided. Ultimately, the result is only as good as the instructions given. One might think that instead of a programming language, we write "instruction language" or prompts, but we still need to be able to make clear to the artificial intelligence what we expect from it. The learning curve for prompt writing may prove to be easier than learning a programming language, but coding skills help guide artificial intelligence more quickly in the right direction.

In programming, a problem is broken down into smaller parts, and these small parts are written in code that the computer can understand. This skill is also very necessary when instructing language models to write code. Even if models improve over time, the instruction "Write a program that can plan activities" may mean different things to different people (and artificial intelligence). Rather, specific requirements and guidelines need to be described to ensure the result is as expected. This does not indicate a weakness in artificial intelligence. No matter how smart and skilled these intelligent systems become, a person (or client) must be able to describe exactly what needs to be created.

Artificial intelligence allows a developer to write smaller pieces much faster and more efficiently. The right assembly of smaller blocks ultimately yields results. We need people who can validate that the created smaller and larger pieces of software are correct. This requires knowledge of problem-solving, programming, and testing. The entire application creation lifecycle actually involves even more – needs must be mapped, the created application must be tested later, and improvements and additions must be made based on feedback.

The leap starts from schools 

In order to support the development of AI engineers, diverse preparation must be consciously emphasized. Compared to today, those seeking work in IT must make conscious choices and place more emphasis on being able to collaborate with artificial intelligence. Deeper knowledge of programming and project management is important, as well as communication skills, analytical ability, adaptability, and flexibility. Since the entire field is developing very rapidly, the AI engineer must be prepared for this as well.

I believe that university education must prepare people for this. It is very important to provide a strong technical foundation along with the ability to be creative in order to utilize one's strengths together with AI. That is why we opened a bachelor's program at the Technical University where, alongside computer science, students can study in depth the fundamentals of developing and managing artificial intelligence.

Artificial intelligence is undoubtedly such a breakthrough technology that its rapid adoption in all areas of life gives the entire country a competitive advantage and leads to the developmental leap that entrepreneurs, the president, and educational visionaries talk about. Everyone must be part of the AI spring, as schools and universities have an extremely important role in determining whether our lives improve with artificial intelligence, whether we are healthier and happier thanks to artificial intelligence, and whether everyone or just some benefit from the new technology.

This article was first published in TalTech's magazine Mente et Manu.